Extract build_mcfg Part 2
v5 cover Extract build_mcfg Part 2
This patch set tries to generalize MCFG table build process. Several patches are already merged and these two are left for the following reasons:
* conflict with latest upstream
* ACPI_PCI dependency fix
* missed reserved[8] in MCFG
v4->v5:
* ACPI_PCI depends on both ACPI and PCI
* rebase on latest master, adjust arm Kconfig
* miss the reserved[8] of MCFG, add it back
* make sure bios-tables-test all OK
Wei Yang (2):
hw/acpi: Consolidate build_mcfg to pci.c
acpi: pci: use build_append_foo() API to construct MCFG
default-configs/i386-softmmu.mak | 1 + hw/acpi/Kconfig | 4 +++ hw/acpi/Makefile.objs | 1 + hw/acpi/pci.c | 57 ++++++++++++++++++++++++++++++++ hw/arm/Kconfig | 1 + hw/arm/virt-acpi-build.c | 17 ---------- hw/i386/acpi-build.c | 18 +--------- include/hw/acpi/acpi-defs.h | 18 ---------- include/hw/acpi/pci.h | 1 + 9 files changed, 66 insertions(+), 52 deletions(-) create mode 100644 hw/acpi/pci.c
Signature · 1 line
2.19.1
PATCH v5 1/2 hw/acpi: Consolidate build_mcfg to pci.c
Now we have two identical build_mcfg functions.
Consolidate them in acpi/pci.c.
---
v5:
* ACPI_PCI depends on both ACPI and PCI
* rebase on latest master, adjust arm Kconfig
v3:
* adjust changelog based on Igor's suggestion
--- default-configs/i386-softmmu.mak | 1 + hw/acpi/Kconfig | 4 +++ hw/acpi/Makefile.objs | 1 + hw/acpi/pci.c | 46 ++++++++++++++++++++++++++++++++ hw/arm/Kconfig | 1 + hw/arm/virt-acpi-build.c | 17 ------------ hw/i386/acpi-build.c | 18 +------------ include/hw/acpi/pci.h | 1 + 8 files changed, 55 insertions(+), 34 deletions(-) create mode 100644 hw/acpi/pci.c
default-configs/i386-softmmu.mak +1 -0
@@ -25,3 +25,4 @@ CONFIG_ISAPC=y CONFIG_I440FX=y CONFIG_Q35=y +CONFIG_ACPI_PCI=y
hw/acpi/Kconfig +4 -0
@@ -23,6 +23,10 @@ config ACPI_NVDIMM bool depends on ACPI +config ACPI_PCI + bool + depends on ACPI && PCI + config ACPI_VMGENID bool default y
hw/acpi/Makefile.objs +1 -0
@@ -11,6 +11,7 @@ common-obj-$(call lnot,$(CONFIG_ACPI_X86)) += acpi-stub.o common-obj-y += acpi_interface.o common-obj-y += bios-linker-loader.o common-obj-y += aml-build.o +common-obj-$(CONFIG_ACPI_PCI) += pci.o common-obj-$(CONFIG_TPM) += tpm.o common-obj-$(CONFIG_IPMI) += ipmi.o
hw/acpi/pci.c new file +46 -0
46 lines not on this page · raw message
hw/arm/Kconfig +1 -0
@@ -19,6 +19,7 @@ config ARM_VIRT select PLATFORM_BUS select SMBIOS select VIRTIO_MMIO + select ACPI_PCI config CHEETAH bool
hw/arm/virt-acpi-build.c +0 -17
@@ -546,23 +546,6 @@ build_srat(GArray *table_data, BIOSLinker *linker, VirtMachineState *vms) "SRAT", table_data->len - srat_start, 3, NULL, NULL); } -static void -build_mcfg (GArray *table_data,BIOSLinker *linker,AcpiMcfgInfo *info) -{ -AcpiTableMcfg *mcfg; -int len =sizeof (*mcfg) +sizeof (mcfg->allocation[0 ]); - - mcfg =acpi_data_push (table_data, len); - mcfg->allocation[0 ].address =cpu_to_le64 (info->base); - -/* Only a single allocation so no need to play with segments */ - mcfg->allocation[0 ].pci_segment =cpu_to_le16 (0 ); - mcfg->allocation[0 ].start_bus_number =0 ; - mcfg->allocation[0 ].end_bus_number =PCIE_MMCFG_BUS (info->size -1 ); - -build_header (linker, table_data, (void *)mcfg,"MCFG" , len,1 ,NULL ,NULL ); -} -/* GTDT */ static void build_gtdt (GArray *table_data,BIOSLinker *linker,VirtMachineState *vms)
hw/i386/acpi-build.c +1 -17
@@ -2405,22 +2405,6 @@ build_srat(GArray *table_data, BIOSLinker *linker, MachineState *machine) table_data->len - srat_start, 1, NULL, NULL); } -static void -build_mcfg_q35 (GArray *table_data,BIOSLinker *linker,AcpiMcfgInfo *info) -{ -AcpiTableMcfg *mcfg; -int len =sizeof (*mcfg) +1 *sizeof (mcfg->allocation[0 ]); - - mcfg =acpi_data_push (table_data, len); - mcfg->allocation[0 ].address =cpu_to_le64 (info->base); -/* Only a single allocation so no need to play with segments */ - mcfg->allocation[0 ].pci_segment =cpu_to_le16 (0 ); - mcfg->allocation[0 ].start_bus_number =0 ; - mcfg->allocation[0 ].end_bus_number =PCIE_MMCFG_BUS (info->size -1 ); - -build_header (linker, table_data, (void *)mcfg,"MCFG" , len,1 ,NULL ,NULL ); -} -/* * VT-d spec 8.1 DMA Remapping Reporting Structure * (version Oct. 2014 or later)
@@ -2690,7 +2674,7 @@ void acpi_build(AcpiBuildTables *tables, MachineState *machine) }if (acpi_get_mcfg (&mcfg)) {acpi_add_table (table_offsets, tables_blob); -build_mcfg_q35 (tables_blob, tables->linker, &mcfg); +build_mcfg (tables_blob, tables->linker, &mcfg); }if (x86_iommu_get_default ()) {IommuType IOMMUType =x86_iommu_get_type ();
include/hw/acpi/pci.h +1 -0
@@ -30,4 +30,5 @@ typedef struct AcpiMcfgInfo { uint32_t size; } AcpiMcfgInfo; +void build_mcfg (GArray *table_data,BIOSLinker *linker,AcpiMcfgInfo *info);#endif
Signature · 1 line
2.19.1
Re: [Qemu-arm] [PATCH v5 1/2] hw/acpi: Consolidate build_mcfg to pci.c
Quoted 6 lines · Wei Yang
Now we have two identical
build_mcfgfunctions.Consolidate them in
acpi/pci.c.Signed-off-by Wei Yang <[email protected]>
Quoted 7 lines
---
v5:
*ACPI_PCIdepends on both ACPI and PCI
* rebase on latest master, adjust armKconfig
v3:
* adjust changelog based on Igor's suggestion
---
same as 2/2 - do not use two --- separators pls.
Quoted 187 lines
default-configs/i386-softmmu.mak | 1 + hw/acpi/Kconfig | 4 +++ hw/acpi/Makefile.objs | 1 + hw/acpi/pci.c | 46 ++++++++++++++++++++++++++++++++ hw/arm/Kconfig | 1 + hw/arm/virt-acpi-build.c | 17 ------------ hw/i386/acpi-build.c | 18 +------------ include/hw/acpi/pci.h | 1 + 8 files changed, 55 insertions(+), 34 deletions(-) create mode 100644 hw/acpi/pci.cdefault-configs/i386-softmmu.mak +1
-0@@ -25,3 +25,4 @@ CONFIG_ISAPC=y CONFIG_I440FX=y CONFIG_Q35=y +CONFIG_ACPI_PCI=yhw/acpi/Kconfig +4
-0@@ -23,6 +23,10 @@ config ACPI_NVDIMM bool depends on ACPI +config ACPI_PCI + bool + depends on ACPI && PCI + config ACPI_VMGENID bool default yhw/acpi/Makefile.objs +1
-0@@ -11,6 +11,7 @@ common-obj-$(call lnot,$(CONFIG_ACPI_X86)) += acpi-stub.o common-obj-y += acpi_interface.o common-obj-y += bios-linker-loader.o common-obj-y += aml-build.o +common-obj-$(CONFIG_ACPI_PCI) += pci.o common-obj-$(CONFIG_TPM) += tpm.o common-obj-$(CONFIG_IPMI) += ipmi.ohw/acpi/pci.c new file +46
-0@@ -0,0 +1,46 @@ +/* +* Support for generating PCI related ACPI tables and passing them to Guests +* +* Copyright (C) 2006 Fabrice Bellard +* Copyright (C) 2008-2010 Kevin O'Connor <[email protected]> +* Copyright (C) 2013-2019 Red Hat Inc +* Copyright (C) 2019 Intel Corporation +* +* Author: Wei Yang <[email protected]> +* Author: Michael S. Tsirkin <[email protected]> +* +* This program is free software; you can redistribute it and/or modify +* it under the terms of the GNU General Public License as published by +* the Free Software Foundation; either version 2 of the License, or +* (at your option) any later version. + +* This program is distributed in the hope that it will be useful, +* but WITHOUT ANY WARRANTY; without even the implied warranty of +* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +* GNU General Public License for more details. + +* You should have received a copy of the GNU General Public License along +* with this program; if not, see <http://www.gnu.org/licenses/>. +*/ + +#include "qemu/osdep.h" +#include "hw/acpi/aml-build.h" +#include "hw/acpi/pci.h" +#include "hw/pci/pcie_host.h" + +void build_mcfg (GArray *table_data,BIOSLinker *linker,AcpiMcfgInfo *info) +{ +AcpiTableMcfg *mcfg; +int len =sizeof (*mcfg) +sizeof (mcfg->allocation[0 ]); + + mcfg =acpi_data_push (table_data, len); + mcfg->allocation[0 ].address =cpu_to_le64 (info->base); + +/* Only a single allocation so no need to play with segments */ + mcfg->allocation[0 ].pci_segment =cpu_to_le16 (0 ); + mcfg->allocation[0 ].start_bus_number =0 ; + mcfg->allocation[0 ].end_bus_number =PCIE_MMCFG_BUS (info->size -1 ); + +build_header (linker, table_data, (void *)mcfg,"MCFG" , len,1 ,NULL ,NULL ); +} +hw/arm/Kconfig +1
-0@@ -19,6 +19,7 @@ config ARM_VIRT select PLATFORM_BUS select SMBIOS select VIRTIO_MMIO + select ACPI_PCI config CHEETAH boolhw/arm/virt-acpi-build.c +0
-17@@ -546,23 +546,6 @@ build_srat(GArray *table_data, BIOSLinker *linker, VirtMachineState *vms) "SRAT", table_data->len - srat_start, 3, NULL, NULL); } -static void -build_mcfg (GArray *table_data,BIOSLinker *linker,AcpiMcfgInfo *info) -{ -AcpiTableMcfg *mcfg; -int len =sizeof (*mcfg) +sizeof (mcfg->allocation[0 ]); - - mcfg =acpi_data_push (table_data, len); - mcfg->allocation[0 ].address =cpu_to_le64 (info->base); - -/* Only a single allocation so no need to play with segments */ - mcfg->allocation[0 ].pci_segment =cpu_to_le16 (0 ); - mcfg->allocation[0 ].start_bus_number =0 ; - mcfg->allocation[0 ].end_bus_number =PCIE_MMCFG_BUS (info->size -1 ); - -build_header (linker, table_data, (void *)mcfg,"MCFG" , len,1 ,NULL ,NULL ); -} -/* GTDT */ static void build_gtdt (GArray *table_data,BIOSLinker *linker,VirtMachineState *vms)hw/i386/acpi-build.c +1
-17@@ -2405,22 +2405,6 @@ build_srat(GArray *table_data, BIOSLinker *linker, MachineState *machine) table_data->len - srat_start, 1, NULL, NULL); } -static void -build_mcfg_q35 (GArray *table_data,BIOSLinker *linker,AcpiMcfgInfo *info) -{ -AcpiTableMcfg *mcfg; -int len =sizeof (*mcfg) +1 *sizeof (mcfg->allocation[0 ]); - - mcfg =acpi_data_push (table_data, len); - mcfg->allocation[0 ].address =cpu_to_le64 (info->base); -/* Only a single allocation so no need to play with segments */ - mcfg->allocation[0 ].pci_segment =cpu_to_le16 (0 ); - mcfg->allocation[0 ].start_bus_number =0 ; - mcfg->allocation[0 ].end_bus_number =PCIE_MMCFG_BUS (info->size -1 ); - -build_header (linker, table_data, (void *)mcfg,"MCFG" , len,1 ,NULL ,NULL ); -} -/* * VT-d spec 8.1 DMA Remapping Reporting Structure * (version Oct. 2014 or later) @@ -2690,7 +2674,7 @@ void acpi_build(AcpiBuildTables *tables, MachineState *machine) }if (acpi_get_mcfg (&mcfg)) {acpi_add_table (table_offsets, tables_blob); -build_mcfg_q35 (tables_blob, tables->linker, &mcfg); +build_mcfg (tables_blob, tables->linker, &mcfg); }if (x86_iommu_get_default ()) {IommuType IOMMUType =x86_iommu_get_type ();include/hw/acpi/pci.h +1
-0@@ -30,4 +30,5 @@ typedef struct AcpiMcfgInfo { uint32_t size; } AcpiMcfgInfo; +void build_mcfg (GArray *table_data,BIOSLinker *linker,AcpiMcfgInfo *info);#endif Signature · 1 line
2.19.1
Re: [Qemu-arm] [Qemu-devel] [PATCH v5 1/2] hw/acpi: Consolidate build_mcfg to pci.c
Quoted 16 lines · On Mon, 20 May 2019 23:37:46 -0400 "Michael S. Tsirkin" <[email protected]> wrote:
Quoted 6 lines · Wei Yang
Now we have two identical
build_mcfgfunctions.Consolidate them in
acpi/pci.c.Signed-off-by Wei Yang <[email protected]>Quoted 7 lines
---
v5:
*ACPI_PCIdepends on both ACPI and PCI
* rebase on latest master, adjust armKconfig
v3:
* adjust changelog based on Igor's suggestion
---
Quoted 1 line
same as 2/2 - do not use two --- separators pls.
I don't really get this requirement, it's common practice on list(s) and used to work just fine.
The 1st separator is added by hand when editing commit message to separate change log from main commit message so that 'git am' would drop change log. While the second separator is added automatically by git when patch is created.
PATCH v5 2/2 acpi: pci: use build_append_foo() API to construct MCFG
In reply to Wei Yangbuild_append_foo() API doesn't need explicit endianness conversions which eliminates a source of errors and it makes build_mcfg() look like declarative definition of MCFG table in ACPI spec, which makes it easy to review.
---
v5:
* miss the reserved[8] of MCFG in last version, add it back
* drop SOBs and make sure bios-tables-test all OK
--- hw/acpi/pci.c | 35 +++++++++++++++++++++++------------ include/hw/acpi/acpi-defs.h | 18 ------------------ 2 files changed, 23 insertions(+), 30 deletions(-)
hw/acpi/pci.c +23 -12
@@ -30,17 +30,28 @@void build_mcfg (GArray *table_data,BIOSLinker *linker,AcpiMcfgInfo *info) { -AcpiTableMcfg *mcfg; -int len =sizeof (*mcfg) +sizeof (mcfg->allocation[0 ]); - - mcfg =acpi_data_push (table_data, len); - mcfg->allocation[0 ].address =cpu_to_le64 (info->base); - -/* Only a single allocation so no need to play with segments */ - mcfg->allocation[0 ].pci_segment =cpu_to_le16 (0 ); - mcfg->allocation[0 ].start_bus_number =0 ; - mcfg->allocation[0 ].end_bus_number =PCIE_MMCFG_BUS (info->size -1 ); - -build_header (linker, table_data, (void *)mcfg,"MCFG" , len,1 ,NULL ,NULL ); +int mcfg_start = table_data->len; + +acpi_data_push (table_data,sizeof (AcpiTableHeader)); + +/* +* PCI Firmware Specification, Revision 3.0 +* 4.1.2 MCFG Table Description. +*/ +/* Reserved */ +build_append_int_noprefix (table_data,0 ,8 ); +/* Base address, processor-relative */ +build_append_int_noprefix (table_data, info->base,8 ); +/* PCI segment group number */ +build_append_int_noprefix (table_data,0 ,2 ); +/* Starting PCI Bus number */ +build_append_int_noprefix (table_data,0 ,1 ); +/* Final PCI Bus number */ +build_append_int_noprefix (table_data,PCIE_MMCFG_BUS (info->size -1 ),1 ); +/* Reserved */ +build_append_int_noprefix (table_data,0 ,4 ); + +build_header (linker, table_data, (void *)(table_data->data + mcfg_start), +"MCFG" , table_data->len - mcfg_start,1 ,NULL ,NULL ); }
include/hw/acpi/acpi-defs.h +0 -18
@@ -449,24 +449,6 @@ struct AcpiSratProcessorGiccAffinity {typedef struct AcpiSratProcessorGiccAffinity AcpiSratProcessorGiccAffinity ; -/* PCI fw r3.0 MCFG table. */ -/* Subtable */ -struct AcpiMcfgAllocation { -uint64_t address;/* Base address, processor-relative */ -uint16_t pci_segment;/* PCI segment group number */ -uint8_t start_bus_number;/* Starting PCI Bus number */ -uint8_t end_bus_number;/* Final PCI Bus number */ -uint32_t reserved; -}QEMU_PACKED ; -typedef struct AcpiMcfgAllocation AcpiMcfgAllocation ; - -struct AcpiTableMcfg { -ACPI_TABLE_HEADER_DEF ; -uint8_t reserved[8 ]; -AcpiMcfgAllocation allocation[0 ]; -}QEMU_PACKED ; -typedef struct AcpiTableMcfg AcpiTableMcfg ; -/* * TCPA Description Table *
Signature · 1 line
2.19.1
Re: [Qemu-arm] [PATCH v5 2/2] acpi: pci: use build_append_foo() API to construct MCFG
Quoted 9 lines · Wei Yang
build_append_foo()API doesn't need explicit endianness conversions which eliminates a source of errors and it makesbuild_mcfg()look like declarative definition of MCFG table in ACPI spec, which makes it easy to review.Signed-off-by Wei Yang <[email protected]>Suggested-by Igor Mammedov <[email protected]>Reviewed-by Igor Mammedov <[email protected]>
Quoted 5 lines
---
v5:
* miss the reserved[8] of MCFG in last version, add it back
* drop SOBs and make sure bios-tables-test all OK
---
Please do not add two --- separators. It breaks git am.
One --- should come after the commit log. Anything after that and until diff is ignored anyway.
Quoted 80 lines
hw/acpi/pci.c | 35 +++++++++++++++++++++++------------ include/hw/acpi/acpi-defs.h | 18 ------------------ 2 files changed, 23 insertions(+), 30 deletions(-)hw/acpi/pci.c +23
-12@@ -30,17 +30,28 @@void build_mcfg (GArray *table_data,BIOSLinker *linker,AcpiMcfgInfo *info) { -AcpiTableMcfg *mcfg; -int len =sizeof (*mcfg) +sizeof (mcfg->allocation[0 ]); - - mcfg =acpi_data_push (table_data, len); - mcfg->allocation[0 ].address =cpu_to_le64 (info->base); - -/* Only a single allocation so no need to play with segments */ - mcfg->allocation[0 ].pci_segment =cpu_to_le16 (0 ); - mcfg->allocation[0 ].start_bus_number =0 ; - mcfg->allocation[0 ].end_bus_number =PCIE_MMCFG_BUS (info->size -1 ); - -build_header (linker, table_data, (void *)mcfg,"MCFG" , len,1 ,NULL ,NULL ); +int mcfg_start = table_data->len; + +acpi_data_push (table_data,sizeof (AcpiTableHeader)); + +/* +* PCI Firmware Specification, Revision 3.0 +* 4.1.2 MCFG Table Description. +*/ +/* Reserved */ +build_append_int_noprefix (table_data,0 ,8 ); +/* Base address, processor-relative */ +build_append_int_noprefix (table_data, info->base,8 ); +/* PCI segment group number */ +build_append_int_noprefix (table_data,0 ,2 ); +/* Starting PCI Bus number */ +build_append_int_noprefix (table_data,0 ,1 ); +/* Final PCI Bus number */ +build_append_int_noprefix (table_data,PCIE_MMCFG_BUS (info->size -1 ),1 ); +/* Reserved */ +build_append_int_noprefix (table_data,0 ,4 ); + +build_header (linker, table_data, (void *)(table_data->data + mcfg_start), +"MCFG" , table_data->len - mcfg_start,1 ,NULL ,NULL ); }include/hw/acpi/acpi-defs.h +0
-18@@ -449,24 +449,6 @@ struct AcpiSratProcessorGiccAffinity {typedef struct AcpiSratProcessorGiccAffinity AcpiSratProcessorGiccAffinity ; -/* PCI fw r3.0 MCFG table. */ -/* Subtable */ -struct AcpiMcfgAllocation { -uint64_t address;/* Base address, processor-relative */ -uint16_t pci_segment;/* PCI segment group number */ -uint8_t start_bus_number;/* Starting PCI Bus number */ -uint8_t end_bus_number;/* Final PCI Bus number */ -uint32_t reserved; -}QEMU_PACKED ; -typedef struct AcpiMcfgAllocation AcpiMcfgAllocation ; - -struct AcpiTableMcfg { -ACPI_TABLE_HEADER_DEF ; -uint8_t reserved[8 ]; -AcpiMcfgAllocation allocation[0 ]; -}QEMU_PACKED ; -typedef struct AcpiTableMcfg AcpiTableMcfg ; -/* * TCPA Description Table * Signature · 1 line
2.19.1
Re: [Qemu-arm] [PATCH v5 2/2] acpi: pci: use build_append_foo() API to construct MCFG
Quoted 10 lines · On Mon, 20 May 2019 23:37:22 -0400 "Michael S. Tsirkin" <[email protected]> wrote:
Quoted 9 lines · Wei Yang
build_append_foo()API doesn't need explicit endianness conversions which eliminates a source of errors and it makesbuild_mcfg()look like declarative definition of MCFG table in ACPI spec, which makes it easy to review.Signed-off-by Wei Yang <[email protected]>Suggested-by Igor Mammedov <[email protected]>Reviewed-by Igor Mammedov <[email protected]>
Quoted 8 lines
Quoted 5 lines
---
v5:
* miss the reserved[8] of MCFG in last version, add it back
* drop SOBs and make sure bios-tables-test all OK
---Please do not add two --- separators. It breaks git am.
that worked just fine for the last xx years and still works for me.
Quoted 3 lines
One --- should come after the commit log. Anything after that and until diff is ignored anyway.
That's fine if we wish to commit change log into history, but
Typically we don't want change log to be committed (as it's useless from history pov), hence we put it after separator to get it dropped on applying.
Quoted 81 lines
Quoted 81 lines
hw/acpi/pci.c | 35 +++++++++++++++++++++++------------ include/hw/acpi/acpi-defs.h | 18 ------------------ 2 files changed, 23 insertions(+), 30 deletions(-)hw/acpi/pci.c +23
-12@@ -30,17 +30,28 @@void build_mcfg (GArray *table_data,BIOSLinker *linker,AcpiMcfgInfo *info) { -AcpiTableMcfg *mcfg; -int len =sizeof (*mcfg) +sizeof (mcfg->allocation[0 ]); - - mcfg =acpi_data_push (table_data, len); - mcfg->allocation[0 ].address =cpu_to_le64 (info->base); - -/* Only a single allocation so no need to play with segments */ - mcfg->allocation[0 ].pci_segment =cpu_to_le16 (0 ); - mcfg->allocation[0 ].start_bus_number =0 ; - mcfg->allocation[0 ].end_bus_number =PCIE_MMCFG_BUS (info->size -1 ); - -build_header (linker, table_data, (void *)mcfg,"MCFG" , len,1 ,NULL ,NULL ); +int mcfg_start = table_data->len; + +acpi_data_push (table_data,sizeof (AcpiTableHeader)); + +/* +* PCI Firmware Specification, Revision 3.0 +* 4.1.2 MCFG Table Description. +*/ +/* Reserved */ +build_append_int_noprefix (table_data,0 ,8 ); +/* Base address, processor-relative */ +build_append_int_noprefix (table_data, info->base,8 ); +/* PCI segment group number */ +build_append_int_noprefix (table_data,0 ,2 ); +/* Starting PCI Bus number */ +build_append_int_noprefix (table_data,0 ,1 ); +/* Final PCI Bus number */ +build_append_int_noprefix (table_data,PCIE_MMCFG_BUS (info->size -1 ),1 ); +/* Reserved */ +build_append_int_noprefix (table_data,0 ,4 ); + +build_header (linker, table_data, (void *)(table_data->data + mcfg_start), +"MCFG" , table_data->len - mcfg_start,1 ,NULL ,NULL ); }include/hw/acpi/acpi-defs.h +0
-18@@ -449,24 +449,6 @@ struct AcpiSratProcessorGiccAffinity {typedef struct AcpiSratProcessorGiccAffinity AcpiSratProcessorGiccAffinity ; -/* PCI fw r3.0 MCFG table. */ -/* Subtable */ -struct AcpiMcfgAllocation { -uint64_t address;/* Base address, processor-relative */ -uint16_t pci_segment;/* PCI segment group number */ -uint8_t start_bus_number;/* Starting PCI Bus number */ -uint8_t end_bus_number;/* Final PCI Bus number */ -uint32_t reserved; -}QEMU_PACKED ; -typedef struct AcpiMcfgAllocation AcpiMcfgAllocation ; - -struct AcpiTableMcfg { -ACPI_TABLE_HEADER_DEF ; -uint8_t reserved[8 ]; -AcpiMcfgAllocation allocation[0 ]; -}QEMU_PACKED ; -typedef struct AcpiTableMcfg AcpiTableMcfg ; -/* * TCPA Description Table * Signature · 1 line
2.19.1
Michael S. Tsirkin
Reviewed-by Igor MammedovRe: [Qemu-arm] [PATCH v5 2/2] acpi: pci: use build_append_foo() API to construct MCFG
Quoted 20 lines · Igor Mammedov
Quoted 17 lines · On Mon, 20 May 2019 23:37:22 -0400 "Michael S. Tsirkin" <[email protected]> wrote:
Quoted 14 lines · Wei Yang
build_append_foo()API doesn't need explicit endianness conversions which eliminates a source of errors and it makesbuild_mcfg()look like declarative definition of MCFG table in ACPI spec, which makes it easy to review.Signed-off-by Wei Yang <[email protected]>Suggested-by Igor Mammedov <[email protected]>Reviewed-by Igor Mammedov <[email protected]>---
v5:
* miss the reserved[8] of MCFG in last version, add it back
* drop SOBs and make sure bios-tables-test all OK
---
Quoted 2 lines
Quoted 1 line
Please do not add two --- separators. It breaks git am.
that worked just fine for the last xx years and still works for me.
Because you don't use git am?
It's true I often am forced to edit your patches to apply them. Pls do not do this it makes no sense.
Quoted 6 lines
Quoted 3 lines
One --- should come after the commit log. Anything after that and until diff is ignored anyway.
That's fine if we wish to commit change log into history, but
Typically we don't want change log to be committed (as it's useless from history pov), hence we put it after separator to get it dropped on applying.
Then you should do
build_append_foo() API doesn't need explicit endianness conversions which eliminates a source of errors and it makes build_mcfg() look like declarative definition of MCFG table in ACPI spec, which makes it easy to review.
---
v5:
* miss the reserved[8] of MCFG in last version, add it back
* drop SOBs and make sure bios-tables-test all OK
hw/acpi/pci.c | 35 +++++++++++++++++++++++------------ include/hw/acpi/acpi-defs.h | 18 ------------------ 2 files changed, 23 insertions(+), 30 deletions(-)
hw/acpi/pci.c +0 -0
Anything between --- and diff is ignored. Second --- is poointless and harmful.
Quoted 81 lines
Quoted 81 lines
Quoted 81 lines
hw/acpi/pci.c | 35 +++++++++++++++++++++++------------ include/hw/acpi/acpi-defs.h | 18 ------------------ 2 files changed, 23 insertions(+), 30 deletions(-)hw/acpi/pci.c +23
-12@@ -30,17 +30,28 @@void build_mcfg (GArray *table_data,BIOSLinker *linker,AcpiMcfgInfo *info) { -AcpiTableMcfg *mcfg; -int len =sizeof (*mcfg) +sizeof (mcfg->allocation[0 ]); - - mcfg =acpi_data_push (table_data, len); - mcfg->allocation[0 ].address =cpu_to_le64 (info->base); - -/* Only a single allocation so no need to play with segments */ - mcfg->allocation[0 ].pci_segment =cpu_to_le16 (0 ); - mcfg->allocation[0 ].start_bus_number =0 ; - mcfg->allocation[0 ].end_bus_number =PCIE_MMCFG_BUS (info->size -1 ); - -build_header (linker, table_data, (void *)mcfg,"MCFG" , len,1 ,NULL ,NULL ); +int mcfg_start = table_data->len; + +acpi_data_push (table_data,sizeof (AcpiTableHeader)); + +/* +* PCI Firmware Specification, Revision 3.0 +* 4.1.2 MCFG Table Description. +*/ +/* Reserved */ +build_append_int_noprefix (table_data,0 ,8 ); +/* Base address, processor-relative */ +build_append_int_noprefix (table_data, info->base,8 ); +/* PCI segment group number */ +build_append_int_noprefix (table_data,0 ,2 ); +/* Starting PCI Bus number */ +build_append_int_noprefix (table_data,0 ,1 ); +/* Final PCI Bus number */ +build_append_int_noprefix (table_data,PCIE_MMCFG_BUS (info->size -1 ),1 ); +/* Reserved */ +build_append_int_noprefix (table_data,0 ,4 ); + +build_header (linker, table_data, (void *)(table_data->data + mcfg_start), +"MCFG" , table_data->len - mcfg_start,1 ,NULL ,NULL ); }include/hw/acpi/acpi-defs.h +0
-18@@ -449,24 +449,6 @@ struct AcpiSratProcessorGiccAffinity {typedef struct AcpiSratProcessorGiccAffinity AcpiSratProcessorGiccAffinity ; -/* PCI fw r3.0 MCFG table. */ -/* Subtable */ -struct AcpiMcfgAllocation { -uint64_t address;/* Base address, processor-relative */ -uint16_t pci_segment;/* PCI segment group number */ -uint8_t start_bus_number;/* Starting PCI Bus number */ -uint8_t end_bus_number;/* Final PCI Bus number */ -uint32_t reserved; -}QEMU_PACKED ; -typedef struct AcpiMcfgAllocation AcpiMcfgAllocation ; - -struct AcpiTableMcfg { -ACPI_TABLE_HEADER_DEF ; -uint8_t reserved[8 ]; -AcpiMcfgAllocation allocation[0 ]; -}QEMU_PACKED ; -typedef struct AcpiTableMcfg AcpiTableMcfg ; -/* * TCPA Description Table * Signature · 1 line
2.19.1
Re: [Qemu-arm] [PATCH v5 2/2] acpi: pci: use build_append_foo() API to construct MCFG
Quoted 21 lines · On Tue, 21 May 2019 10:34:03 -0400 "Michael S. Tsirkin" <[email protected]> wrote:
Quoted 20 lines · Igor Mammedov
Quoted 17 lines · On Mon, 20 May 2019 23:37:22 -0400 "Michael S. Tsirkin" <[email protected]> wrote:
Quoted 14 lines · Wei Yang
build_append_foo()API doesn't need explicit endianness conversions which eliminates a source of errors and it makesbuild_mcfg()look like declarative definition of MCFG table in ACPI spec, which makes it easy to review.Signed-off-by Wei Yang <[email protected]>Suggested-by Igor Mammedov <[email protected]>Reviewed-by Igor Mammedov <[email protected]>---
v5:
* miss the reserved[8] of MCFG in last version, add it back
* drop SOBs and make sure bios-tables-test all OK
---
Quoted 6 lines
Quoted 2 lines
Quoted 1 line
Please do not add two --- separators. It breaks git am.
that worked just fine for the last xx years and still works for me.
Because you don't use git am?
It's true I often am forced to edit your patches to apply them. Pls do not do this it makes no sense.
I do not use it as much as you, but I do use git am to test patches from list and it works with double separator as expected (the second separator is ignored).
Quoted 7 lines
Quoted 5 lines
Quoted 2 lines
One --- should come after the commit log. Anything after that and until diff is ignored anyway.
That's fine if we wish to commit change log into history, but
Typically we don't want change log to be committed (as it's useless from history pov), hence we put it after separator to get it dropped on applying.Then you should do
the problem is that the second separator is added by git, so one would have to edit it out after format-patch (which is error prone in case of multi-patch series).
I don't know much about git though, so could you suggest how to handle situation to avoid the second --- (i.e. how do you do it)?
PS:
There is 'git notes' to add/track per patch annotations, but it a bit tedious to use notes compared to in-lining change log in commit message.
Quoted 108 lines
build_append_foo()API doesn't need explicit endianness conversions which eliminates a source of errors and it makesbuild_mcfg()look like declarative definition of MCFG table in ACPI spec, which makes it easy to review.Signed-off-by Wei Yang <[email protected]>Suggested-by Igor Mammedov <[email protected]>Reviewed-by Igor Mammedov <[email protected]>---
v5:
* miss the reserved[8] of MCFG in last version, add it back
* drop SOBs and make sure bios-tables-test all OKhw/acpi/pci.c | 35 +++++++++++++++++++++++------------ include/hw/acpi/acpi-defs.h | 18 ------------------ 2 files changed, 23 insertions(+), 30 deletions(-)hw/acpi/pci.c +0
-0Anything between --- and diff is ignored. Second --- is poointless and harmful.
Quoted 81 lines
Quoted 81 lines
Quoted 81 lines
hw/acpi/pci.c | 35 +++++++++++++++++++++++------------ include/hw/acpi/acpi-defs.h | 18 ------------------ 2 files changed, 23 insertions(+), 30 deletions(-)hw/acpi/pci.c +23
-12@@ -30,17 +30,28 @@void build_mcfg (GArray *table_data,BIOSLinker *linker,AcpiMcfgInfo *info) { -AcpiTableMcfg *mcfg; -int len =sizeof (*mcfg) +sizeof (mcfg->allocation[0 ]); - - mcfg =acpi_data_push (table_data, len); - mcfg->allocation[0 ].address =cpu_to_le64 (info->base); - -/* Only a single allocation so no need to play with segments */ - mcfg->allocation[0 ].pci_segment =cpu_to_le16 (0 ); - mcfg->allocation[0 ].start_bus_number =0 ; - mcfg->allocation[0 ].end_bus_number =PCIE_MMCFG_BUS (info->size -1 ); - -build_header (linker, table_data, (void *)mcfg,"MCFG" , len,1 ,NULL ,NULL ); +int mcfg_start = table_data->len; + +acpi_data_push (table_data,sizeof (AcpiTableHeader)); + +/* +* PCI Firmware Specification, Revision 3.0 +* 4.1.2 MCFG Table Description. +*/ +/* Reserved */ +build_append_int_noprefix (table_data,0 ,8 ); +/* Base address, processor-relative */ +build_append_int_noprefix (table_data, info->base,8 ); +/* PCI segment group number */ +build_append_int_noprefix (table_data,0 ,2 ); +/* Starting PCI Bus number */ +build_append_int_noprefix (table_data,0 ,1 ); +/* Final PCI Bus number */ +build_append_int_noprefix (table_data,PCIE_MMCFG_BUS (info->size -1 ),1 ); +/* Reserved */ +build_append_int_noprefix (table_data,0 ,4 ); + +build_header (linker, table_data, (void *)(table_data->data + mcfg_start), +"MCFG" , table_data->len - mcfg_start,1 ,NULL ,NULL ); }include/hw/acpi/acpi-defs.h +0
-18@@ -449,24 +449,6 @@ struct AcpiSratProcessorGiccAffinity {typedef struct AcpiSratProcessorGiccAffinity AcpiSratProcessorGiccAffinity ; -/* PCI fw r3.0 MCFG table. */ -/* Subtable */ -struct AcpiMcfgAllocation { -uint64_t address;/* Base address, processor-relative */ -uint16_t pci_segment;/* PCI segment group number */ -uint8_t start_bus_number;/* Starting PCI Bus number */ -uint8_t end_bus_number;/* Final PCI Bus number */ -uint32_t reserved; -}QEMU_PACKED ; -typedef struct AcpiMcfgAllocation AcpiMcfgAllocation ; - -struct AcpiTableMcfg { -ACPI_TABLE_HEADER_DEF ; -uint8_t reserved[8 ]; -AcpiMcfgAllocation allocation[0 ]; -}QEMU_PACKED ; -typedef struct AcpiTableMcfg AcpiTableMcfg ; -/* * TCPA Description Table * Signature · 1 line
2.19.1
Re: [Qemu-arm] [PATCH v5 2/2] acpi: pci: use build_append_foo() API to construct MCFG
In reply to Wei YangQuoted 38 lines · Wei Yang
build_append_foo()API doesn't need explicit endianness conversions which eliminates a source of errors and it makesbuild_mcfg()look like declarative definition of MCFG table in ACPI spec, which makes it easy to review.Signed-off-by Wei Yang <[email protected]>Suggested-by Igor Mammedov <[email protected]>Reviewed-by Igor Mammedov <[email protected]>---
v5:
* miss the reserved[8] of MCFG in last version, add it back
* drop SOBs and make sure bios-tables-test all OK--- hw/acpi/pci.c | 35 +++++++++++++++++++++++------------ include/hw/acpi/acpi-defs.h | 18 ------------------ 2 files changed, 23 insertions(+), 30 deletions(-)hw/acpi/pci.c +0
-12@@ -30,17 +30,28 @@void build_mcfg (GArray *table_data,BIOSLinker *linker,AcpiMcfgInfo *info) { -AcpiTableMcfg *mcfg; -int len =sizeof (*mcfg) +sizeof (mcfg->allocation[0 ]); - - mcfg =acpi_data_push (table_data, len); - mcfg->allocation[0 ].address =cpu_to_le64 (info->base); - -/* Only a single allocation so no need to play with segments */ - mcfg->allocation[0 ].pci_segment =cpu_to_le16 (0 ); - mcfg->allocation[0 ].start_bus_number =0 ; - mcfg->allocation[0 ].end_bus_number =PCIE_MMCFG_BUS (info->size -1 ); - -build_header (linker, table_data, (void *)mcfg,"MCFG" , len,1 ,NULL ,NULL );
hw/acpi/pci.c line 42
Quoted 10 lines
hw/acpi/pci.c +10
-0+int mcfg_start = table_data->len; + +acpi_data_push (table_data,sizeof (AcpiTableHeader)); + +/* +* PCI Firmware Specification, Revision 3.0 +* 4.1.2 MCFG Table Description. +*/ +/* Reserved */ +build_append_int_noprefix (table_data,0 ,8 );
below is in fact
Memory Mapped Enhanced Configuration Space Base Address Allocation Structure
maybe document this?
Quoted 46 lines
hw/acpi/pci.c +13
-0+/* Base address, processor-relative */ +build_append_int_noprefix (table_data, info->base,8 ); +/* PCI segment group number */ +build_append_int_noprefix (table_data,0 ,2 ); +/* Starting PCI Bus number */ +build_append_int_noprefix (table_data,0 ,1 ); +/* Final PCI Bus number */ +build_append_int_noprefix (table_data,PCIE_MMCFG_BUS (info->size -1 ),1 ); +/* Reserved */ +build_append_int_noprefix (table_data,0 ,4 ); + +build_header (linker, table_data, (void *)(table_data->data + mcfg_start), +"MCFG" , table_data->len - mcfg_start,1 ,NULL ,NULL ); }include/hw/acpi/acpi-defs.h +0
-18@@ -449,24 +449,6 @@ struct AcpiSratProcessorGiccAffinity {typedef struct AcpiSratProcessorGiccAffinity AcpiSratProcessorGiccAffinity ; -/* PCI fw r3.0 MCFG table. */ -/* Subtable */ -struct AcpiMcfgAllocation { -uint64_t address;/* Base address, processor-relative */ -uint16_t pci_segment;/* PCI segment group number */ -uint8_t start_bus_number;/* Starting PCI Bus number */ -uint8_t end_bus_number;/* Final PCI Bus number */ -uint32_t reserved; -}QEMU_PACKED ; -typedef struct AcpiMcfgAllocation AcpiMcfgAllocation ; - -struct AcpiTableMcfg { -ACPI_TABLE_HEADER_DEF ; -uint8_t reserved[8 ]; -AcpiMcfgAllocation allocation[0 ]; -}QEMU_PACKED ; -typedef struct AcpiTableMcfg AcpiTableMcfg ; -/* * TCPA Description Table * Signature · 1 line
2.19.1
Re: [Qemu-arm] [PATCH v5 2/2] acpi: pci: use build_append_foo() API to construct MCFG
Quoted 53 lines · Michael S. Tsirkin
Quoted 38 lines · Wei Yang
build_append_foo()API doesn't need explicit endianness conversions which eliminates a source of errors and it makesbuild_mcfg()look like declarative definition of MCFG table in ACPI spec, which makes it easy to review.Signed-off-by Wei Yang <[email protected]>Suggested-by Igor Mammedov <[email protected]>Reviewed-by Igor Mammedov <[email protected]>---
v5:
* miss the reserved[8] of MCFG in last version, add it back
* drop SOBs and make sure bios-tables-test all OK--- hw/acpi/pci.c | 35 +++++++++++++++++++++++------------ include/hw/acpi/acpi-defs.h | 18 ------------------ 2 files changed, 23 insertions(+), 30 deletions(-)hw/acpi/pci.c +0
-12@@ -30,17 +30,28 @@void build_mcfg (GArray *table_data,BIOSLinker *linker,AcpiMcfgInfo *info) { -AcpiTableMcfg *mcfg; -int len =sizeof (*mcfg) +sizeof (mcfg->allocation[0 ]); - - mcfg =acpi_data_push (table_data, len); - mcfg->allocation[0 ].address =cpu_to_le64 (info->base); - -/* Only a single allocation so no need to play with segments */ - mcfg->allocation[0 ].pci_segment =cpu_to_le16 (0 ); - mcfg->allocation[0 ].start_bus_number =0 ; - mcfg->allocation[0 ].end_bus_number =PCIE_MMCFG_BUS (info->size -1 ); - -build_header (linker, table_data, (void *)mcfg,"MCFG" , len,1 ,NULL ,NULL );hw/acpi/pci.c line 42
Quoted 10 lines
hw/acpi/pci.c +10
-0+int mcfg_start = table_data->len; + +acpi_data_push (table_data,sizeof (AcpiTableHeader)); + +/* +* PCI Firmware Specification, Revision 3.0 +* 4.1.2 MCFG Table Description. +*/ +/* Reserved */ +build_append_int_noprefix (table_data,0 ,8 );below is in fact
Memory Mapped Enhanced Configuration Space Base Address Allocation Structure
Quoted 2 lines
maybe document this?
Sure. Let me add this.
Quoted 46 lines
Quoted 46 lines
hw/acpi/pci.c +13
-0+/* Base address, processor-relative */ +build_append_int_noprefix (table_data, info->base,8 ); +/* PCI segment group number */ +build_append_int_noprefix (table_data,0 ,2 ); +/* Starting PCI Bus number */ +build_append_int_noprefix (table_data,0 ,1 ); +/* Final PCI Bus number */ +build_append_int_noprefix (table_data,PCIE_MMCFG_BUS (info->size -1 ),1 ); +/* Reserved */ +build_append_int_noprefix (table_data,0 ,4 ); + +build_header (linker, table_data, (void *)(table_data->data + mcfg_start), +"MCFG" , table_data->len - mcfg_start,1 ,NULL ,NULL ); }include/hw/acpi/acpi-defs.h +0
-18@@ -449,24 +449,6 @@ struct AcpiSratProcessorGiccAffinity {typedef struct AcpiSratProcessorGiccAffinity AcpiSratProcessorGiccAffinity ; -/* PCI fw r3.0 MCFG table. */ -/* Subtable */ -struct AcpiMcfgAllocation { -uint64_t address;/* Base address, processor-relative */ -uint16_t pci_segment;/* PCI segment group number */ -uint8_t start_bus_number;/* Starting PCI Bus number */ -uint8_t end_bus_number;/* Final PCI Bus number */ -uint32_t reserved; -}QEMU_PACKED ; -typedef struct AcpiMcfgAllocation AcpiMcfgAllocation ; - -struct AcpiTableMcfg { -ACPI_TABLE_HEADER_DEF ; -uint8_t reserved[8 ]; -AcpiMcfgAllocation allocation[0 ]; -}QEMU_PACKED ; -typedef struct AcpiTableMcfg AcpiTableMcfg ; -/* * TCPA Description Table * Signature · 1 line
2.19.1
Signature · 2 lines
Wei Yang Help you, Help me